[ANN] intel's netdev-ci for i40e

[ANN] intel's netdev-ci for i40e

[Pielech, Adrian]

Hi folks,

I'm pleased to announce that we have set up infrastructure for testing
netdev on our e700 NIC (i40e), which extends our current engagement with
e810 NICs(ice).

This work plugs into netdev-ci initiative by netdev maintainers to run
kselftests against current proposed net-next-hw branch on real hardware.

Our results are here:
https://netdev-ci-results.intel.com/i40e-results/results.json

Human friendly results browser:
https://netdev-ci-results.intel.com/results_browser/index.html?drv=i40e

Adrian

Re: [ANN] intel's netdev-ci for i40e

[Jakub Kicinski]

On Tue, 3 Feb 2026 13:35:11 +0100 Pielech, Adrian wrote:
> Hi folks,
> 
> I'm pleased to announce that we have set up infrastructure for testing
> netdev on our e700 NIC (i40e), which extends our current engagement with
> e810 NICs(ice).
> 
> This work plugs into netdev-ci initiative by netdev maintainers to run
> kselftests against current proposed net-next-hw branch on real hardware.
> 
> Our results are here:
> https://netdev-ci-results.intel.com/i40e-results/results.json
> 
> Human friendly results browser:
> https://netdev-ci-results.intel.com/results_browser/index.html?drv=i40e

Excellent!

Looks like the GRO test result parsing is off, the test cases are
reported as for example:

	gro-test-lro-ipv6-ip-tos # XFAIL Device does not support...

Looks like this affects all your runners, I haven't noticed earlier.

Re: [ANN] intel's netdev-ci for i40e

[Pielech, Adrian]

On 2/4/2026 4:08 AM, Jakub Kicinski wrote:
> On Tue, 3 Feb 2026 13:35:11 +0100 Pielech, Adrian wrote:
>> Hi folks,
>>
>> I'm pleased to announce that we have set up infrastructure for testing
>> netdev on our e700 NIC (i40e), which extends our current engagement with
>> e810 NICs(ice).
>>
>> This work plugs into netdev-ci initiative by netdev maintainers to run
>> kselftests against current proposed net-next-hw branch on real hardware.
>>
>> Our results are here:
>> https://netdev-ci-results.intel.com/i40e-results/results.json
>>
>> Human friendly results browser:
>> https://netdev-ci-results.intel.com/results_browser/index.html?drv=i40e
> 
> Excellent!
> 
> Looks like the GRO test result parsing is off, the test cases are
> reported as for example:
> 
> 	gro-test-lro-ipv6-ip-tos # XFAIL Device does not support...
> 
> Looks like this affects all your runners, I haven't noticed earlier.

I have fixed that, XFAIL should be reported as fail even if further
message suggests that the test should report SKIP due to "device does
not support...".

Historical results were parsed again to remove test cases:
"[...] # XFAIL [...]".

Re: [ANN] intel's netdev-ci for i40e

[Jakub Kicinski]

On Wed, 4 Feb 2026 13:15:03 +0100 Pielech, Adrian wrote:
> On 2/4/2026 4:08 AM, Jakub Kicinski wrote:
> > Looks like the GRO test result parsing is off, the test cases are
> > reported as for example:
> > 
> > 	gro-test-lro-ipv6-ip-tos # XFAIL Device does not support...
> > 
> > Looks like this affects all your runners, I haven't noticed earlier.  
> 
> I have fixed that, XFAIL should be reported as fail even if further
> message suggests that the test should report SKIP due to "device does
> not support...".
> 
> Historical results were parsed again to remove test cases:
> "[...] # XFAIL [...]".

Looks good now, thanks!

One more note, the "driver metadata" is currently reported as:

{"devlink":{"driver":"i40e", ...

looks like you have some extra nests there. That's perfectly fine,
but we need there to be a "driver" key in the outmost object.
Right now our UIs show "undefined" for the drivers here:
https://netdev.bots.linux.dev/devices.html

Re: [ANN] intel's netdev-ci for i40e

[Pielech, Adrian]

On 2/5/2026 2:40 AM, Jakub Kicinski wrote:
> On Wed, 4 Feb 2026 13:15:03 +0100 Pielech, Adrian wrote:
>> On 2/4/2026 4:08 AM, Jakub Kicinski wrote:
>>> Looks like the GRO test result parsing is off, the test cases are
>>> reported as for example:
>>>
>>> 	gro-test-lro-ipv6-ip-tos # XFAIL Device does not support...
>>>
>>> Looks like this affects all your runners, I haven't noticed earlier.
>>
>> I have fixed that, XFAIL should be reported as fail even if further
>> message suggests that the test should report SKIP due to "device does
>> not support...".
>>
>> Historical results were parsed again to remove test cases:
>> "[...] # XFAIL [...]".
> 
> Looks good now, thanks!
> 
> One more note, the "driver metadata" is currently reported as:
> 
> {"devlink":{"driver":"i40e", ...
> 
> looks like you have some extra nests there. That's perfectly fine,
> but we need there to be a "driver" key in the outmost object.
> Right now our UIs show "undefined" for the drivers here:
> https://netdev.bots.linux.dev/devices.html

I have moved position of "driver" from device.devlink.driver to
device.driver. Regenerated current and past results. You can refresh
your dashboard as well to wipe duplicated entries.

Re: [ANN] intel's netdev-ci for i40e

[Jakub Kicinski]

On Wed, 11 Feb 2026 10:17:52 +0100 Pielech, Adrian wrote:
> > Looks good now, thanks!
> > 
> > One more note, the "driver metadata" is currently reported as:
> > 
> > {"devlink":{"driver":"i40e", ...
> > 
> > looks like you have some extra nests there. That's perfectly fine,
> > but we need there to be a "driver" key in the outmost object.
> > Right now our UIs show "undefined" for the drivers here:
> > https://netdev.bots.linux.dev/devices.html  
> 
> I have moved position of "driver" from device.devlink.driver to
> device.driver. Regenerated current and past results. You can refresh
> your dashboard as well to wipe duplicated entries.

Please don't regenerate past results in the future.
It's a pain to have to clean up the DB.